AI Video API Cost in 2026: List Price vs Usable-Clip Cost
Per-second rates are the smallest term in a video bill. Attempts per usable clip went from 2-3 to 7-8 this year.
Faceless YouTube Channel Automation: $0.30 per Short, Start to Finish
One key, four steps: script model, three 5-second clips, ffmpeg concat. 15 seconds of vertical video for $0.3009 in 260 seconds, and 3 parts that break.
Seedance 2.5 Image-to-Video: First and Last Frame Control (2026)
Send both ends and Seedance 2.5 lands on them: closing-frame error 5.4 against 35.9 for 2.0 Mini. Setting aspect_ratio is a 400. Billed $0.11/s at 480p.
"Transparent Background Is Not Supported for This Model": 3 Fixes
gpt-image-2 returns this 400 while gpt-image-1.5 returns a real RGBA PNG. Five image models on one endpoint: one works, one errors, three return opaque.
Video Generation API Polling: 202, polling_url, Wait Times (2026)
POST returns 202 in 0.8s, then nothing for 83 to 253 seconds. Eight identical 5s jobs, a 3x spread. Cancel fails once it runs, and the URL dies in 24h.
DeepSeek V4 Flash Vision: Images Cap at 384 Tokens
DeepSeek's vision variant of V4 Flash costs the same as the text model. Images cap at 384 input tokens, and thinking mode adds 80 input tokens per call.
Discounted LLM APIs: 13 Models Below List, GPT-5.6 Sol at Half Price
Thirteen models across five series bill below list. Which cuts are the vendor's own promotion, and which are ofox's, including Sol at half OpenAI's list.
Gemini 3.7 Flash API: $0.75 Pricing, minimal Returns 400
Gemini 3.7 Flash bills $0.75 in / $3.75 out, same as 3.6 Flash. reasoning_effort minimal now returns 400, and the cheapest legal tier costs 8.4x more.
Opus 5 vs Grok 4.6 Cost: 3.6x the Bill for 1.75x the Code
One real task, measured: Opus 5 costs 3.6x Grok 4.6 and returns 1.75x as much code, in half the wall clock. Plus the usage field that hides the bill.
GLM 5.3 API: Pricing, Endpoints, and reasoning_effort
The GLM 5.3 API costs $1.4 in / $4.4 out, ships three base URLs, and defaults reasoning_effort to max.